For researchers and programmatic clients
Research data and machine interface
Schema 1.0. These static, cacheable files are generated from the same public historical-person model as the human catalogue. Start with the compact resolver index, then fetch one dossier or depth-one family network.
Predictable endpoints
| Purpose | Path |
|---|---|
| Compact global index | /data/people-index.json |
| Person dossier | /people/<lowercase-wikitree-id>.json |
| Depth-one network | /people/<lowercase-wikitree-id>.network.json |
| Exact normalised-name candidates | /data/resolve/<name-slug>.json |
The static resolver returns candidates, never a genealogical conclusion. Filter or score candidates using birth year, all locations, spouse names, parents, children and occupations in people-index.json. The generated score is deterministic: normalised name, birth-year tolerance, location and spouse matches add explainable reasons; conflicts subtract a documented amount.
Claims, evidence and relationships
The dossier follows Person → Claim → Evidence → Source. Stable hash-based IDs prevent rebuilds from renumbering unchanged objects. Canonical claim statuses are unknown, possible, probable, strongly_supported, proved, disputed and contradicted; the original project label is retained in status_label. Source quality is original, derivative, secondary, tree_only or unknown.
A relationship with tree_relationship: true says the exported WikiTree currently connects the people. It is not documentary proof. Children with relationship_source: reconstructed_parent_reference were recovered by reversing child profiles’ Father/Mother fields. A proved relationship carries claim and evidence IDs from the controlled evidence assessment.
Research questions and leads
open_questions come from versioned local research cases. research_leads include pending case targets plus conservatively extracted profile-note actions only when they contain a concrete archive reference, URL, or a record type with a constrained year. Repository, collection/reference, date bounds and reason are retained where supplied. No archive references are invented.
Provenance and interpretation
- Mapped associations, WikiTree tree links, extracted citations, controlled relationship assessments and local research findings retain separate provenance labels.
independence_groupgroups citations that resolve to the same underlying URL or source key.- Estimated dates and locations remain explicitly labelled.
- A cluster is a research grouping, not proof of kinship.
- Full profile captures remain available in
wikitree-profile-evidence.json, but compact dossiers omit raw biographies.